Inline
On 22/11/2009, at 5:03 PM, David E Jones wrote:


On Nov 21, 2009, at 8:40 PM, Scott Gray wrote:

Hi David,

I've looked at doing that a couple of times now but it usually ends up presenting a few difficulties for something that is seemingly simple: 1. You're stuck with either using entities or view entities, I guess you could use a predefined list but that would defeat the purpose somewhat. My example below is a little boring but one possibly common example of where you might need more power is an autocomplete for a party name that could be either person or party group where the results of two entities need to be combined. You could call a service to retrieve the list in my example.

With most things that is the idea of the form widget, ie provide an easy way to do the most common things (like lookup based on an entity or view-entity, using any conditions specified plus the text entered), and then have facilities for dropping down to scripting or services or templates or whatever in order to be able to handle anything so that the less common cases are still quite doable. Right now the autocomplete implementation supports the less common cases well, but is inefficient for the more simple and frequent stuff.

As far as I know the existing auto-completer (on-field-event-update-area right?) remains unused OOTB, which I think may be a testament to how useful it is.
Agreed. The main reason for me was that I had to code also the event separately, which means double work.


2. There is no way of indicating what field you actually want to search against.

This would typically be a search on whatever the description is made up of (ie that's what users expect).
Searching on one field is not useful for most of the cases. For example to search for a party, it is good to search in partyId, firstName, middleName, lastName, groupName fields. With other entities it would be good to search at lease in ID and description fields.

What about a case like this:
        <field name="currencyUomId" title="${uiLabelMap.CommonCurrency}">
<drop-down allow-empty="true" no-current-selected-key="${defaultOrganizationPartyCurrencyUomId}"> <entity-options key-field-name="uomId" description="${description} - ${abbreviation}" entity-name="Uom"> <entity-constraint name="uomTypeId" operator="equals" value="CURRENCY_MEASURE"/>
                    <entity-order-by field-name="description"/>
                </entity-options>
            </drop-down>
        </field>
The abbreviation is useful to display but not necessarily something you want to search against. And even if you did want to search against it how would you parse this string for a search?


3. There seems to be a common trend with auto-completers that they can be paired with an additional lookup form to allow for an advanced search, adding this capability to the drop down element only makes sense for an auto-completer.

Not sure what you mean by this... but it sounds interesting. Do you mean something like a multi-field auto-completer?

I mean you use a lookup form (like we do now with the lookup element) as a fall back when the user wants to do a more advanced search than what an auto-completer allows.
+ 1



In the end it seemed to me that while a drop-down and an auto-completer are quite similar, the differences between the two were enough to warrant a separate element (plus it's a damn sight easier :-)

What you propose might be easier to implement, but wouldn't it be pretty similar to what exists now... with the difference being you can specify the event inline instead of coding it separately?

Well I'm saying it is easier to implement that an auto-complete within a drop-down. In a way I guess it is similar to the on-field-event-update-area element, but it is actually part of a larger POC I've been working on from time to time whereby every field element can have a field-events sub-element, whenever the event occurs the actions are run and the result is sent back to the form. The main differences are that you don't specify a target (because the event is inline) and you don't specify an update area. The client-side takes the json result and updates all necessary fields according to a standard client-side form model.


Some examples of values you might choose to return:
(Substitute fieldName with the actual field-name)
fieldName.value (this could update a text input's value, a drop-down selection, a hidden input value etc.) fieldName.options (a list of value/description maps for updating a drop-down's options e.g. select a country and the state list gets repopulated) fieldName.error (an error message perhaps after some server side validation) fieldName.disabled (boolean to disable a field e.g. perhaps a submit button until some error is resolved)
fieldGroupId.hidden (boolean to show or hide a field group)

I still don't have solid proposal in place really, just an idea of what it might be nice to be able to support and possible way of achieving it. I only started discussing the auto-complete portion because of Bilgin's POC, I thought I better get my 2 cents in before things headed in a different direction. But yes the main differences between this and on-field-event-update-area is the inline actions and the ability to update multiple "areas".

I liked very much the idea of inline events and the direction of this conversation, but as for the lookup autocompleter (which is the only target of my POC code), I think my solution fits betters. There are more than five hundred lookup fields in the project, and adding autocompleter code to each field(entity name to search in, fields to search, field(s) to return as result, there might be other options as well) plus the existing lookup code seems not feasible and would clutter the code. Also in lots of forms, there are more than one references to the same lookup (party name lookup) and this will end up in repeating code for the same autocompleter.

On the other hand my solution doesn't require any change on the form field definitions. There is already a target specified for the lookup, and event (screen). So by adding 2 lines per lookup screen, it is possible to return a properly formated response to the autocompleter. And it will be more consistent to specify all the autocomplete options for a spcific lookup in one place.

Even the autocompleter extension seems trivial, it reduces significantly the clicks while working with forms. If you want to see, I could commit it for a day, only for testing and then revert.

Regards,
Bilgin

Reply via email to